Gensyn

gensyn.ai · Compute · 1 record · JSON

Gensyn Compute

fails: quotaemailno cardphoneDwrite-gated

docs.gensyn.ai/testnet/rl-swarm/getting-started · GPU cloud · compute/gensyn · JSON · compare

Free tier	No consumer-facing capacity. Quoted: 'RL Swarm is a peer-to-peer reinforcement-learning swarm you can run on a laptop or a GPU server', with system requirements 'A 64-bit arm64 or x86 CPU with at least 32 GB RAM, or an officially supported NVIDIA GPU (3090, 4090, 5090, A100, H100)'. The dashboard states: 'There are no official swarms running right now.' unit GiB; metered on wall-clock; kind not stated
Price	none stated
Per-object cap	not applicable
Retention	not applicable
Egress	not stated
Identity	email a Hugging Face write token: the getting-started page is quoted as requiring 'a Hugging Face write token', which in turn implies a Hugging Face account that issues it. No card, hold, phone or ID is stated, and the email kind rests on that Hugging Face account rather than on a quoted signup form. Debit and prepaid do not arise because no payment method is mentioned. The 32 GB RAM or listed NVIDIA GPU requirement is a capacity gate, not an identity one.
Limits	write-gated The direction is inverted: Gensyn pays suppliers who bring hardware, so it consumes the one thing the client does not have. Even as a supplier lane it fails first on the 32 GB RAM or supported-GPU floor, then on 'no official swarms running right now'. source The direction is inverted: Gensyn pays suppliers who bring hardware, so it consumes the one thing the client does not have. Even as a supplier lane it fails first on the 32 GB RAM or supported-GPU floor, then on 'no official swarms running right now'.
Binding	write-gated The direction is inverted: Gensyn pays suppliers who bring hardware, so it consumes the one thing the client does not have. Even as a supplier lane it fails first on the 32 GB RAM or supported-GPU floor, then on 'no official swarms running right now'.
Interface	unknown inferred from the tooling the provider documents; not probed
Tooling	not stated
Composes with	None. Gensyn is a market for selling compute, not buying it, and it has no live swarm. A chain that includes Gensyn would need the client to own the GPU.
Reliability	no reliability event recorded
Notes	REFUTATION OF A COMMON MISREADING. Gensyn appears in free-GPU roundups because RL Swarm is free to join, but joining means contributing hardware, not receiving capacity. Three independent facts kill it for a user: it needs 32 GB RAM or a listed NVIDIA card, its dashboard says no official swarms are running, and its reward unit is a token allocation out of a 2 percent testnet pool with no published per-node formula. A third-party guide also records that renting a GPU to run a Gensyn node costs USD 65 to USD 300 per month, which is negative margin for a client with no budget.
Evidence rung	documented
Checked	2026-10-22 (live); stale on 2026-11-21
Parsed numbers	`$.limits[0].value` limit:write-gated 32.0 GB

Evidence (4)

date	source	what it proved
2026-10-22	docs.gensyn.ai/testnet/rl-swarm/getting-started	quoted: requires a 64-bit arm64/x86 CPU with at least 32 GB RAM or a supported NVIDIA GPU (3090, 4090, 5090, A100, H100); a Hugging Face write token is required
2026-10-22	dashboard.gensyn.ai/	quoted: 'There are no official swarms running right now. Please check back later'; the network reports 427,530 on-chain users
2026-10-22	docs.gensyn.ai/testnet	public testnet launched March 2025; the network tracks participation and makes payments, it does not rent capacity to consumers
2026-10-22	depinrun.com/guides/gensyn-node-operator-guide	third-party operator guide states 'No compute-provider earnings have been published for Gensyn mainnet' and that testnet rewards were 'relative and pool-based'

More fields (1)

{
 "wave2": {
  "class": "decentralized-training-network",
  "free_tier_kind": null,
  "shape": {
   "vcpu": null,
   "ram_gib": 32,
   "gpu_class": null,
   "gpu_count": null
  }
 }
}

Unknown or withheld values (9)

$.egress.metered egress-unknown
$.egress.terms not-stated
$.interface interface-unknown
$.per_object_cap.notes process
$.per_object_cap.stated not-applicable
$.reliability_note none-observed
$.retention.notes process
$.retention.stated not-applicable
$.tooling not-stated